Download Report

Application Report
SLAA668 – March 2015
Designing an Ultra-Low-Power (ULP) Application With
MSP432™ Microcontrollers
Dung Dang .............................................................................................................. MSP Applications
Atul Lele ................................................................................. Architect – Senior Member Technical Staff
ABSTRACT
With the growing system complexity in ultra-low-power microcontroller applications, minimizing the overall
energy consumption is one of the most difficult problems to solve. Multiple aspects including silicon, other
onboard hardware components, and application software must be considered. There are some obvious
generic techniques that can be used to reduce energy consumption such as reducing operating voltage or
frequency. Many of these generic techniques may not greatly reduce energy consumption independently,
but taken as a whole, the results can be significant, as there are many interdependencies across these
components.
This application report offers an overview of various low-power features of the industry's leading ultra-lowpower microcontroller and, more importantly, how to exploit optimal combinations of these features to
achieve the lowest energy consumption for a given application. This application report uses the industrystandard ULPBench™ benchmark from EEMBC as a case study to map individual and combinations of
low-power features on the MSP432™ platform to a relevant application scenario.
1
2
3
4
5
6
7
8
9
Contents
Power Optimization Vectors on a Microcontroller ....................................................................... 2
Power System ................................................................................................................ 3
Clock System ................................................................................................................. 5
Memory Execution Optimization ........................................................................................... 5
Optimizing Power of an Entire MCU System............................................................................. 8
Summary .................................................................................................................... 10
Other Methods to Optimize Device Power Consumption ............................................................. 10
Disclaimer ................................................................................................................... 11
References .................................................................................................................. 11
List of Figures
1
DC-DC and LDO Regulator: Features, Advantages, and Tradeoffs .................................................. 4
2
MSP432P4xx ULPBench Results
3
MSP432P4xx ULPBench Results
4
......................................................................................... 8
......................................................................................... 9
MSP432P4xx ULPBench Results ........................................................................................ 10
List of Tables
1
LDO and DCDC Current Consumption ................................................................................... 3
2
Effects of Increasing System Operating Frequency ..................................................................... 8
MSP432, MSP430, EnergyTrace, LaunchPad are trademarks of Texas Instruments.
ULPBench is a trademark of Embedded Microprocessor Benchmarking Consortium Embedded Microprocessor Benchmarking Consortium.
All other trademarks are the property of their respective owners.
SLAA668 – March 2015
Submit Documentation Feedback
Designing an Ultra-Low-Power (ULP) Application With MSP432™
Microcontrollers
Copyright © 2015, Texas Instruments Incorporated
1
Power Optimization Vectors on a Microcontroller
1
www.ti.com
Power Optimization Vectors on a Microcontroller
The MSP432 microcontrollers integrate a number of power enhancements that include lower active power
consumption, low-power (LP) mode current consumption, and more efficient peripherals with lower current
consumption. In addition, the MSP432 devices also provide a number of options and power configurations
that enable developers to further optimize the power consumption for a specific MSP432 application.
Nonetheless, for any given microcontroller, there are a number of vectors that can be used to optimize the
power consumption of the device.
[Vector 1] Reduce operating voltage
Power is the product of voltage and current. For a given application system that consumes a certain
amount of current, lower supply voltage can help reduce the power consumption, with the constraint that
minimum voltage requirement is met. This requirement can sometimes be related to the microcontroller
supply voltage level itself (for MSP432P4xx minimum VCC = 1.62 V) or the setting the internal voltage
regular to appropriate voltage level. It could be dependent on the CPU operating frequency or a minimum
voltage required by a certain peripheral to be fully functional. For example, if a 2.5-V reference voltage is
required by the ADC in the application, supply voltage must be greater than or equal to 2.5 V.
[Vector 2] Reduce the operating frequency
As can be deduced from any microcontroller data sheet, power and current consumption are directly
proportional to the operation frequency. In many scenarios, higher operating frequency means that the
CPU can execute code and complete the task faster. However, in certain real-time scenarios, many
application activities are timing-dependent or event-driven. Having the CPU running faster in an idle loop
waiting for a certain event to trigger or waiting for serial data to come in at a lower baud rate can consume
additional power that can be saved. For these scenarios, either putting the device into a low-power mode
or reducing the operating frequency might be worth investigating.
[Vector 3] Maximize sleep time
A typical low-power embedded application spends most of its lifetime in two modes: active mode
executing meaningful tasks and low-power mode with minimal activity other than time-keeping or waiting
for an interrupt or event to wake up to active mode. Low-power mode can consumes much less current
than active mode—on the MSP432P4xx family, LPM3 current can be as low as approximately 750 nA
while active mode current can be up to several milliamps. Minimizing active time and maximizing time
spent in low-power modes can significantly reduce the overall current consumption. This can usually be
achieved by minimizing the active task, optimizing the active code, or increasing the operating speed in
active mode.
[Vector 4] Minimize transition time
In addition to time allocated for active and low-power modes, some applications might unknowingly spend
a considerable amount of time transitioning between these various power modes. If the transitions times
go unnoticed, they might contribute to the total energy consumption increase. One of the power
optimization activities is to identify all of the system transitions and determine if any of them can be
reduced or removed.
[Vector 5] Solving the intermodule dependencies
The previous vectors are all possible options to individually optimize. However, on a given microcontroller
platforms, these vectors might have some interdependencies, or optimizing on a particular vector might
negatively affect another. For example, reducing operating frequency could potentially increase active duty
cycle. Therefore, it is even more important to take into account the intermodule dependencies and
determine optimal combinations of settings for a given platform.
The remaining sections of this application report describe the system peripherals of MSP432 MCUs,
identify critical features and options, correlate them back to one or more of the five vectors above, and
determine how they can help to optimize power consumption. The key core peripherals on MSP432P4xx
devices include: power system [Power Controller Module (PCM)], clock system, and memory system
(flash, SRAM, and ROM). More importantly, the intermodule dependencies specific to MSP432 are
discussed to help show the combinations that makes sense for an MSP432 application.
2
Designing an Ultra-Low-Power (ULP) Application With MSP432™
Microcontrollers
Copyright © 2015, Texas Instruments Incorporated
SLAA668 – March 2015
Submit Documentation Feedback
Power System
www.ti.com
2
Power System
Many MCUs scale performance with voltage; as the voltage drops, the MCU core operates at a reduced
frequency. This can increase energy consumption if the MCU must stay awake longer to perform an
equivalent amount of work. With MSP432 MCUs, however, the core is able to operate at relatively good
speed at the lowest voltage. Thus, operating at a lower voltage gives a true and full improvement in
energy efficiency. [Vector 1]
2.1
CPU and Digital Logic Core Voltage
The MSP432P4xx family of devices requires a secondary core voltage (VCORE) for its internal digital
operation in addition to the primary one applied to the device (VCC). In general, VCORE supplies the CPU,
memories (flash and RAM), and other digital modules, while VCC (connected to the DVCC and AVCC pins
of the device) supplies I/Os and all analog modules (including the oscillators). The VCORE output is
maintained using a dedicated voltage regulator. VCORE is programmable with two predefined voltage levels,
and each level is restricted to particular maximum operating frequency. This provides programmable
dynamic voltage frequency scaling (DVFS) operation.
This directly correlates to [Vector 1], because the digital logic of the system is now supplied with a lower
voltage, it consumes less power than on a microcontroller where the entire system operates out of the
higher VCC supply rail.
Furthermore, if the application operating frequency does not require the higher VCORE voltage, reducing the
VCORE voltage level further reduces the power consumption of the device. For example, specifically for
MSP432P4xx family, if maximum operating clock frequency is ≤24 MHz, the VCORE0 option can be
chosen, which is 1.2 V (typical) compared to VCORE1 which is 1.4 V (typical). VCORE0 provides better
power consumption at a given frequency (≤24 MHz) than VCORE1.
2.2
Regulators: LDO and DCDC
To generate the secondary core voltage (VCORE) from the primary one (DVCC), the MSP432P4xx family
uses two regulators: a low-dropout voltage regulator (LDO) as the default regulator, and an inductor-based
DC-to-DC step-down switching regulator (DC-DC) as a secondary or optional one. The LDO and DC-DC
are connected in parallel. The two modules share a common output (the digital supply VCORE rail) and a
common reference or target voltage.
For the majority of microcontrollers with multiple voltage rails (VCC and VCORE), the LDO regulator is
typically used due to a number of advantages such as cost-effective, relatively noise-free output due to no
switching component, faster to ramp up and down from low-power modes, among others. These
advantages might be critical for certain application with stringent requirements related analog or RF
performance or cost-sensitivity. However, the LDO regulator might not be ideal for power savings,
because current drawn by supply is same as that of the current drawn from LDO: Iin = Iload = Ivcore.
On the other hand, the DC-DC regulator's main advantage is the significant power saving compared to the
LDO. Unlike a linear regulator, which generates a voltage rail from another by dropping the extra voltage
across a linear, effectively passive, element (thereby wasting power equivalent to the voltage dropped
multiplied by the load current drawn), the DC-DC efficiently generates one voltage rail from another using
active (inductor and capacitor) elements. In an ideal DC-DC, power drawn from the output rail (VCORE ×
Iload) equals the power drawn from the input rail (VCC × Iin). Iin is therefore smaller than Iload by the fraction
VCORE / VCC. After accounting for efficiency losses, the DC-DC typically has efficiency in the range of 75%
to 90% for moderate to maximum loads. This allows the DC-DC regulator to provide significant power
saving of up to 45% compared to the LDO regulator. Table 1 shows the current consumption comparison
between the DC-DC and LDO when the system operates at 24 MHz and 48 MHz.
Table 1. LDO and DCDC Current Consumption
Regulator
CPU Frequency = 24 MHz
CPU Frequency = 48 MHz
LDO
3950 µA
7600 µA
DC-DC
2200 µA
4600 µA
SLAA668 – March 2015
Submit Documentation Feedback
Designing an Ultra-Low-Power (ULP) Application With MSP432™
Microcontrollers
Copyright © 2015, Texas Instruments Incorporated
3
Power System
www.ti.com
While the DCDC power saving for active mode is apparent, the LDO's advantages and flexibility might be
required by the application in certain scenarios, as explained previously. The LDO is also the default
regulator that is used upon device startup or after returning to active mode from low-power mode. So any
DC-DC regulator usage should also take into account the transition time from LDO to DC-DC as well as
DC-DC back to LDO prior to going back to low-power mode.
There are several additional advantages/drawbacks for both regulators. Figure 1 summarizes the
advantages and disadvantages of the LDO and DC-DC to help developers determine how and when it is
possible to use the DC-DC regulator to maximize power saving.
Typically, developers must choose between using either an LDO or DC/DC convertor for all use cases. By
providing two options for power regulation, developers can dynamically optimize regulation based on the
current mode of operation. When the system is in a standby mode of operation, for example, the LDO can
be used to minimize wake time. For operating modes or use cases where active current plays a larger role
in power consumption, the DC/DC convertor can be used. For more details on DC-DC usage as well indepth analysis on its advantages and tradeoffs, refer to Maximizing MSP432P4xx Voltage Regulator
Efficiency: DC-DC and LDO Features and Tradeoffs (SLAA640).
Figure 1. DC-DC and LDO Regulator: Features, Advantages, and Tradeoffs
2.3
Low-Power Modes
The MSP432P4xx family provides a number of low-power operating modes that originated in the
MSP430™ devices including LPM3, LPM4, LPM3.5, and LPM4.5. Similar to earlier MSP devices, these
low-power modes on MSP432P4xx consume extremely little power, putting the MSP432P4xx in a great
position for power saving.
In addition to the traditional MSP low-power modes, the MSP432P4xx family introduce another class of
low-power modes that can provide additional power saving: low-frequency power modes. The lowfrequency modes are special low-power low-frequency options that can be used in conjunction with both
active and LPM0 modes. In these low-frequency modes, memory and peripherals can execute at
maximum speed of 128 kHz. Because the regulator can reduce the drive strength to supply minimal
current to drive the entire low-frequency system, the device's total current consumption can be as little as
<80 µA when using these modes.
Refer to Leveraging Low-Frequency Power Modes on MSP432P4xx Microcontrollers for more details on
these special power modes (SLAA657).
4
Designing an Ultra-Low-Power (ULP) Application With MSP432™
Microcontrollers
Copyright © 2015, Texas Instruments Incorporated
SLAA668 – March 2015
Submit Documentation Feedback
Clock System
www.ti.com
3
Clock System
As described in Section 1, power and current consumption is directly proportional to the frequency of
operation (f). Naturally if (f) is reduced, current consumption also reduces linearly. But, reducing frequency
may reduce overall system throughput or may increase the overall energy consumption. The reason is
that the device may have to be in an active operating mode (AM_VCORE0 or AM_VCORE1) for more
time and in low-power mode for less time. Always operating at high frequency to complete an active mode
task also may not help, because of the peak current requirements during such short bursts, which in turn
can affect the battery capacity and hence battery life. Therefore, the optimal selection of clock frequency
for system operation is critical to optimize the energy consumption of the system.
The Clock System (CS) provides options such as choices of different oscillators for different clock
sources, dynamic selection, and clock dividers to enable optimal power consumption
The digitally controlled oscillator (DCO) provides options such as changing frequency on the fly and fine
tuning to an intermediate frequency of 0.98 MHz to 52 MHz, minimal current consumption of <160 µA at
24 MHz and <220 µA at 48 MHz. The maximum DCO clock frequency drift with temperature in external
resistor mode is ±35 ppm/°C, and the drift for internal resistor mode is ±250 ppm/°C. An application can
program the DCO for the best combination of optimal power consumption and required accuracy. If higher
accuracy is needed, the application needs to use the on-chip high-frequency crystal oscillator (HFXT) with
the trade-off being higher current consumption.
The CS also provides flexible clock source and frequency selection for different on-chip clock sources
(MCLK, SMCLK, and ACLK). It can also divide the clocks dynamically, and all of this can be done with
one control register (CSCTL1).
For example, DCO can be used as the main clock source for MCLK and SMCLK, HSMCLK. LFXT can be
used as the clock source for RTC. For slower peripherals such as I2C, the DIVS and DIVHS register bits
can be configured to select slower source clock frequency. For ADC, MODOSC can be used as a quick
ON/OFF oscillator that remains enabled only when ADC conversion operation is ongoing.
The ULPBench example in Section 5 uses the DCO as the main clock source at 24 MHz instead of at
48 MHz for optimal current consumption.
Refer to the CS section in the MSP432P401xx microcontroller technical reference manual, data sheet, and
the application report Multi-Frequency Range and Tunable DCO on MSP432P4xx Microcontrollers
(SLAA658) for further details.
4
Memory Execution Optimization
Although typical code execution involves opcode and literal fetches from memory, it may also involve
continuous and intermittent data writes to and reads from stack (SRAM), function calls from software
library provided in flash or ROM, data writes to or reads from peripherals, DMA data transfers, and other
operations. A significant portion of the power consumption in a typical MCU originates from code
execution from a nonvolatile memory such as flash. Hence, it is beneficial to optimize memory accesses to
reduce as much power consumption as possible during code execution.
4.1
Flash
MSP432P401xx has flash as the main nonvolatile memory for application program code and data.
4.1.1
Wait State and Clock Dependency
Low-power flash memory is typically slower than the high-frequency clocks used at system level such as
CPU or DMA. Therefore, at higher operating frequency, flash memory requires wait-stated accesses,
which means that the CPU is halted for certain number of clock cycles depending on the CPU-to-flash
frequency ratio. Refer to the device data sheet for details.
SLAA668 – March 2015
Submit Documentation Feedback
Designing an Ultra-Low-Power (ULP) Application With MSP432™
Microcontrollers
Copyright © 2015, Texas Instruments Incorporated
5
Memory Execution Optimization
4.1.2
www.ti.com
Data and Instruction Buffers
To offer optimal power consumption and performance across predominantly contiguous memory
accesses, the flash controller offers a read buffering feature. If read buffering is enabled, the flash memory
is always read in entire 128-bit chunks, even though the read access may only be 8, 16, or 32 bits wide.
The 128-bit data and its associated address is buffered by the flash controller, so that subsequent
accesses (expected to be contiguous in nature) within the same 128-bit address boundary are serviced by
the buffer. Using this scheme, the flash accesses have wait states only when the 128-bit boundary is
crossed, while read accesses within the buffer's range are serviced without any bus stalls. If read buffering
is disabled, accesses to the flash bypasses the buffer and the data read from the flash is limited to the
width of the access (8, 16, or 32 bits). Each bank has independent settings for the read buffer. In addition,
within each bank, the application has the flexibility of enabling read buffering either for instruction fetches
only, for data fetches only, or for both.
4.1.3
Flash Power Consumption
Flash power consumption is dependent on the number of bits read. If 128 bits are read, power consumed
is higher than if 32 bits are read but not four times the power of 32-bit read power. So for contiguous or
linear code execution or data fetches, TI recommends enabling the buffer to enhance performance and
reduce power consumption.
4.1.4
•
•
•
•
•
4.1.5
What Options to Choose for Code Execution From Flash
Enabling buffer definitely gives performance advantage but whether it can help for getting better power
consumption or not depends on the linear or nonlinear execution of the code or data reads. If an
application needs higher throughput always at the cost of current consumption, buffers can always be
enabled. With wait-states >0 configured for flash, buffers are recommended to be always enabled for
higher throughput.
In general, TI recommends enabling instruction and data buffers if the code execution or data reads is
in linear fashion to prevent the controller from constantly accessing flash and ultimately reduce to
reduce overall current consumption (refer to Section 4.1.2).
– Specific example is when the code is running with 0-wait states, there is no performance or
throughput advantage. In that case, the code profile really determines whether to enable the buffer
(with a power advantage for linear code execution) or to disable the buffer.
Because the buffer configuration is dynamic, it can be enabled or disabled for specific code sections or
function calls. If there is a function with significant size with mostly linear code execution, TI
recommends enabling and disabling the buffer dynamically before and after the function call
respectively.
It is also recommended to check the current consumption for your specific application and determine
specific combinations of settings that work best for that scenario. MSP432P4xx family of devices
support EnergyTrace™ technology, which can enable developers to construct a detailed energy profile
of the system and correlate back to areas in the code with high contribution to energy consumption.
Based on the findings, developers can leverage capabilities provided by MSP432P401xx to
dynamically change wait-states, enabling/disabling buffers which can be chosen to cater to specific
application needs.
Refer to ULPBench results for an example analysis on how and when to use flash buffer.
Flash Access Benchmark Registers (*BMRK* Registers in Flash Controller)
The flash controller offers two counters for application benchmarking purposes. This feature is extremely
useful for monitoring the number of flash accesses. Using these counts, the application can manage the
optimal wait-states and the buffer enable or disable settings as discussed in previous sections.
• Instruction fetch benchmark counter (32 bits) – Increments on each instruction fetch to the flash
• Data fetch benchmark counter (32 bits) – Increments on each data fetch to the flash
6
Designing an Ultra-Low-Power (ULP) Application With MSP432™
Microcontrollers
Copyright © 2015, Texas Instruments Incorporated
SLAA668 – March 2015
Submit Documentation Feedback
Memory Execution Optimization
www.ti.com
4.2
SRAM
SRAMs are standard read-write memories on MCUs. Particularly on MSP432P401xx, SRAMs work at the
same speed as the CPU clock frequency so code execution from SRAM clearly gives a performance and
throughput advantage compared to code execution from flash. There is a power consumption advantage
for SRAM compared to flash as well. Power consumption (commonly measured in µA/MHz) for SRAM is
significantly lower than the µA/MHz consumption of flash memory. Therefore, TI recommends executing
small loops or functions that are frequently used from SRAM instead of flash for performance and power
benefits.
This may be handled by using a smart function to copy a function or subset of code from flash to SRAM
and execute from SRAM whenever needed. This SRAM bank can be kept enabled during LPM3 mode so
that the contents are retained and copying the same function over is not needed after LPM3 wakeup.
SRAM retention
• SRAM_BANKRET in the SYS module controls the retention of SRAM contents in LPM3 mode.
Depending on the application, either ALL or a FEW banks are programmed to be retained in LPM3
mode which can affect the current consumption in LPM3.
4.3
ROM
In addition to flash, the MSP432P4xx also provides read-only memory (ROM) as a second nonvolatile
memory option. As the name indicates, the ROM content is programmed when the device is
manufactured, and after that memory contents can only be read or executed but not modified.
MSP432P4xx provides 32Kbytes of on-chip ROM including the built-in peripheral driver library (DriverLib)
application program interfaces (APIs) to provide developers with a highly-abstracted set of APIs to
exercise various functionalities of the MSP432P4xx peripherals. The easy-to-understand ROM DriverLib
APIs allow developers to speed up the process of learning how to configure peripheral registers, change
power modes, and other standard tasks.
There are a number of benefits when using ROM DriverLib. First, the DriverLib in ROM is thoroughly
tested and optimized, and it represents the TI-recommended procedures to exercise the peripherals. In
the context of this application report, executing out of ROM yields both higher performance (0 wait state
access to ROM) and better power consumption (much lower than flash execution, and even better than
SRAM). Both of these performance increases, faster execution time and lower power consumption,
directly result in minimizing the energy consumption of the device. [Vector 3]
Maximizing the driver library use helps to reduce flash memory accesses and enable better power
consumption.
Other than the power benefit, there is an indirect benefit of using software driver library. Leveraging
DriverLib in ROM also frees valuable flash memory space for application code. Shifting this additional
approximately 31Kbytes of memory towards the high-level application code for intelligent software to
uniquely differentiate the application is a much better allocation than taking up the memory for routine
peripheral configuration tasks that, if implemented perfectly, would offer only marginal improvement over
the TI-provided ROM solution.
SLAA668 – March 2015
Submit Documentation Feedback
Designing an Ultra-Low-Power (ULP) Application With MSP432™
Microcontrollers
Copyright © 2015, Texas Instruments Incorporated
7
Optimizing Power of an Entire MCU System
5
www.ti.com
Optimizing Power of an Entire MCU System
Now that we have examined several modules within the MSP432 microcontroller system and how they
can be configured to optimize and reduce power consumption, let's take a look at how these components
can be used together and any potential dependencies or constraints they might have on each other.
ULPBench by EEMBC is used as a case study to understand possible trade-offs.
5.1
Is Faster Always Better?
The MSP432 microcontroller as a whole is a heavily clock-gated system, as the CPU and operating clock
frequency can ultimately have a significant effect not just on the clock power consumption but also how
much power required to sustain other peripherals. As a matter of fact, multiple peripherals' power
consumption can be throttled or boosted depending on the CPU frequency requirement.
Furthermore, the flash memory operation is limited to 12 MHz. Running the CPU at any frequency faster
than 12 MHz requires an appropriate amount of flash wait-states to stall the CPU until the flash memory
content can be completely fetched.
Combining these dependencies, Table 2 summarizes the effects of increasing the system operating
frequency.
Table 2. Effects of Increasing System Operating Frequency
System Frequency
5.2
VCORE
LDO/DC-DC
Flash Wait States
10 kHz to 128 kHz
0
LDO (LF mode)
0
128 kHz to 12 MHz
0
LDO/DC-DC
0
12 MHz to 16 MHz
0
LDO/DC-DC
1
16 MHz to 24 MHz
0
LDO/DC-DC
1
24 MHz to 32 MHz
1
LDO/DC-DC
1
32 MHz to 48 MHz
1
LDO/DC-DC
2
Speed Does Not Always Improve Energy Efficiency
Running the CPU at 48 MHz in can result in a lower ULPMark score than running it at 24 MHz. At
48 MHz, workload is definitely completed faster than at 24 MHz, but core voltage must be increased and
there are mode transition latencies involved that can result in lower ULPMark scores. This is described in
the following sections as part of the effect due to transition latencies. In Figure 2, 3-, 12-, and 24-MHz
scores are with 0 wait states, and 48-MHz scores are with 2 wait states.
180
155.9
ULPMark™-CP
160
158.6
120.5
140
120
107.6
100
108.9
86.09
80
87.79
123.9
155.8
143.5
123.9
113
113
105.2
103.9
91.33
60
3
12
24
48
Operating Frequency (MHz)
LDO_BUF_EN
DCDC_BUF_EN
LDO_BUF_DIS
DCDC_BUF_DIS
Figure 2. MSP432P4xx ULPBench Results
8
Designing an Ultra-Low-Power (ULP) Application With MSP432™
Microcontrollers
Copyright © 2015, Texas Instruments Incorporated
SLAA668 – March 2015
Submit Documentation Feedback
Optimizing Power of an Entire MCU System
www.ti.com
5.3
Efficiency of Flash Instruction and Data Buffers and Correlation to Wait States
128-bit instruction and data buffers for flash reads can be used for better throughput and power
consumption if code execution is linear. If execution is not linear, there may be overhead due to reading
128 bits by using more power and discarding some of the words. MSP432P4xx provides flexibility to
enable or disable the buffers dynamically, and software can optimally use this feature depending on the
linear or nonlinear nature of the code execution (see Figure 3).
The 158.6 score with the buffer disabled is higher, which implies nonlinearity of the code at 0-wait state
condition. But with higher operating frequency and with 1-wait state, buffer disabled gives a better score of
155.8 at 24 MHz.
180
155.9
ULPMark™-CP
160
158.6
140
120
155.8
143.5
123.9
120.5
107.6
123.9
100
108.9
86.09
80
87.79
60
113
113
105.2
103.9
91.33
3
12
24
48
Operating Frequency (MHz)
LDO_BUF_EN
DCDC_BUF_EN
LDO_BUF_DIS
DCDC_BUF_DIS
Figure 3. MSP432P4xx ULPBench Results
5.4
VCORE Switching Latency Diminishes Energy Saving
Code execution of ULPBench workload function during active mode of MSP432P4xx takes insignificant
number of cycles (<0.06%) with high optimization of IAR compiler at 24 MHz, 1-wait state condition
compared the time spent in LPM3 mode (approximately 99.94%). So any additional time it spends in
active mode can severely hamper the ULPBench score.
For example, if the lowest power consumption is needed in LPM3 mode with VCORE0 setting, and the
highest performance is needed at 48 MHz, then there can be many mode transitions involved as per the
guidelines in the MSP432P4xx Technical Reference Manual (SLAU356).
AM_LDO_VCORE0 → AM_LDO_VCORE1 → AM_DCDC_VCORE1 → AM_LDO_VCORE1 →
AM_LDO_VCORE0 → LPM3_VCORE0
For example, additional transitions can be avoided if higher performance is needed as soon as the device
wakes up from LPM3 mode.
AM_LDO_VCORE0 → AM_LDO_VCORE1 → AM_DCDC_VCORE1 → AM_LDO_VCORE1 →
LPM3_VCORE1
In both the cases, transition time can be significant and can affect the duration for which the device is in
active mode, which affects the ULPBench score. So it is important to understand four aspects from the
application's standpoint.
• Duration spent in active operational mode by an application
• Duration spent in low-power mode by an application
• Duration spent in transitioning between different active modes
• Duration spent in transitioning between active and low-power modes
SLAA668 – March 2015
Submit Documentation Feedback
Designing an Ultra-Low-Power (ULP) Application With MSP432™
Microcontrollers
Copyright © 2015, Texas Instruments Incorporated
9
Summary
www.ti.com
Another reason for results not being so attractive for configurations involving these mode transitions is that
the transition functions are written in such a way that the CPU waits for the mode transition to be complete
before the actual code execution starts. For example, in the condition of 24 MHz, 1 wait state, buffer
enabled, the ULPBench score is 155.8. If the LPM3_VCORE0 → AM_LDO_VCORE0 transition function is
changed to initiate the mode transition and continue with the rest of the code execution without waiting for
PMR_BUSY flag to be set, the score can go as high as 167.4.
ULPMark™-CP
180
155.9
160
158.6
120.5
140
120
155.8
143.5
123.9
107.6
123.9
100
108.9
86.09
80
87.79
113
113
105.2
103.9
91.33
60
3
12
24
48
Operating Frequency (MHz)
LDO_BUF_EN
DCDC_BUF_EN
LDO_BUF_DIS
DCDC_BUF_DIS
Figure 4. MSP432P4xx ULPBench Results
6
Summary
Section 1 described generic vectors to reduce power consumption on a generic microcontroller design,
and the rest of the sections described how these can be easily achieved on the MSP432 platform:
• Choose the right combination of CPU frequency and flash wait states for optimal performance. Refer to
the data sheet for additional information.
• Use features that allow 128-bit instruction and data buffer in flash controller to be enabled and disabled
dynamically according to linear or nonlinear nature of code execution from flash memory. This can
affect performance and power consumption.
• Choose optimal operating voltage point to balance performance and power consumption
• Optimize different mode transition latencies by minimizing the transitions depending on the time
duration spent in each mode
7
Other Methods to Optimize Device Power Consumption
7.1
Optimize Device Power Consumption
•
•
•
10
Terminate and configure I/Os as soon as possible after application begins. By default, all of the I/Os
are in input mode. Depending on the application, relevant I/Os should be driven externally, should be
internally or externally pulled up or down, or should be programed in output mode with value driven as
0 or 1.
Use smart module enable and disable features to consume power only when module is needed to be
active. For example, use the UCSWRST bit in eUSCIA and eUSCIB modules to keep the module in
reset or active.
Use an optimal clock source. MODOSC or SYSOSC are also automatically enabled to provide
MODCLK or SYSCLK to ADC14 when needed and are disabled when not needed for ADC14 or for
rest of the device. Use these as ADC14 clock sources for optimal ADC operation.
Designing an Ultra-Low-Power (ULP) Application With MSP432™
Microcontrollers
Copyright © 2015, Texas Instruments Incorporated
SLAA668 – March 2015
Submit Documentation Feedback
Other Methods to Optimize Device Power Consumption
www.ti.com
7.2
Leverage Other MSP432 ULP Tools
The MSP432 microcontroller ecosystem also provides a unique and powerful set of ultra-low-power tools
and software that developers can use to further optimize the application power consumption. Starting with
a well-designed embedded systems using silicon and hardware knowledge such as from this application
report, embedded developers can then take advantage of static code analyzer tool such as ULP Advisor
to further identify areas in the code to optimize for execution time and to reduce power consumption. Next,
during debug phase, developers can use EnergyTrace technology to construct an energy profile of the
application and correlate critical energy consumption spikes to areas in the code. This allows developers
to identify high-energy components in the configuration or software for further power improvement.
8
Disclaimer
•
•
•
9
ULPBench measurements, current consumption measurements were done on limited number of units
which were from a specific manufacturing lot. Results measured on a LaunchPad™ development kit or
with different units may vary.
External board components can also affect current consumption. For example, LFXT current
consumption can change significantly depending on the external capacitors.
Refer to the electrical specification in the data sheet for various parameters and their MIN, TYP, and
MAX values.
References
1.
2.
3.
4.
5.
6.
Multi-Frequency Range and Tunable DCO on MSP432P401xx Microcontrollers (SLAA658)
Maximizing MSP432P4xx Voltage Regulator Efficiency (SLAA640)
Getting Started With EEMBC ULPBench™ on MSP‑EXP432P401R (SLAA667)
MSP432P4xx Technical Reference Manual (SLAU356)
EnergyTrace Technology (www.ti.com/energygtrace)
ULP Advisor (www.ti.com/ulpadvisor)
SLAA668 – March 2015
Submit Documentation Feedback
Designing an Ultra-Low-Power (ULP) Application With MSP432™
Microcontrollers
Copyright © 2015, Texas Instruments Incorporated
11
IMPORTANT NOTICE
Texas Instruments Incorporated and its subsidiaries (TI) reserve the right to make corrections, enhancements, improvements and other
changes to its semiconductor products and services per JESD46, latest issue, and to discontinue any product or service per JESD48, latest
issue. Buyers should obtain the latest relevant information before placing orders and should verify that such information is current and
complete. All semiconductor products (also referred to herein as “components”) are sold subject to TI’s terms and conditions of sale
supplied at the time of order acknowledgment.
TI warrants performance of its components to the specifications applicable at the time of sale, in accordance with the warranty in TI’s terms
and conditions of sale of semiconductor products. Testing and other quality control techniques are used to the extent TI deems necessary
to support this warranty. Except where mandated by applicable law, testing of all parameters of each component is not necessarily
performed.
TI assumes no liability for applications assistance or the design of Buyers’ products. Buyers are responsible for their products and
applications using TI components. To minimize the risks associated with Buyers’ products and applications, Buyers should provide
adequate design and operating safeguards.
TI does not warrant or represent that any license, either express or implied, is granted under any patent right, copyright, mask work right, or
other intellectual property right relating to any combination, machine, or process in which TI components or services are used. Information
published by TI regarding third-party products or services does not constitute a license to use such products or services or a warranty or
endorsement thereof. Use of such information may require a license from a third party under the patents or other intellectual property of the
third party, or a license from TI under the patents or other intellectual property of TI.
Reproduction of significant portions of TI information in TI data books or data sheets is permissible only if reproduction is without alteration
and is accompanied by all associated warranties, conditions, limitations, and notices. TI is not responsible or liable for such altered
documentation. Information of third parties may be subject to additional restrictions.
Resale of TI components or services with statements different from or beyond the parameters stated by TI for that component or service
voids all express and any implied warranties for the associated TI component or service and is an unfair and deceptive business practice.
TI is not responsible or liable for any such statements.
Buyer acknowledges and agrees that it is solely responsible for compliance with all legal, regulatory and safety-related requirements
concerning its products, and any use of TI components in its applications, notwithstanding any applications-related information or support
that may be provided by TI. Buyer represents and agrees that it has all the necessary expertise to create and implement safeguards which
anticipate dangerous consequences of failures, monitor failures and their consequences, lessen the likelihood of failures that might cause
harm and take appropriate remedial actions. Buyer will fully indemnify TI and its representatives against any damages arising out of the use
of any TI components in safety-critical applications.
In some cases, TI components may be promoted specifically to facilitate safety-related applications. With such components, TI’s goal is to
help enable customers to design and create their own end-product solutions that meet applicable functional safety standards and
requirements. Nonetheless, such components are subject to these terms.
No TI components are authorized for use in FDA Class III (or similar life-critical medical equipment) unless authorized officers of the parties
have executed a special agreement specifically governing such use.
Only those TI components which TI has specifically designated as military grade or “enhanced plastic” are designed and intended for use in
military/aerospace applications or environments. Buyer acknowledges and agrees that any military or aerospace use of TI components
which have not been so designated is solely at the Buyer's risk, and that Buyer is solely responsible for compliance with all legal and
regulatory requirements in connection with such use.
TI has specifically designated certain components as meeting ISO/TS16949 requirements, mainly for automotive use. In any case of use of
non-designated products, TI will not be responsible for any failure to meet ISO/TS16949.
Products
Applications
Audio
www.ti.com/audio
Automotive and Transportation
www.ti.com/automotive
Amplifiers
amplifier.ti.com
Communications and Telecom
www.ti.com/communications
Data Converters
dataconverter.ti.com
Computers and Peripherals
www.ti.com/computers
DLP® Products
www.dlp.com
Consumer Electronics
www.ti.com/consumer-apps
DSP
dsp.ti.com
Energy and Lighting
www.ti.com/energy
Clocks and Timers
www.ti.com/clocks
Industrial
www.ti.com/industrial
Interface
interface.ti.com
Medical
www.ti.com/medical
Logic
logic.ti.com
Security
www.ti.com/security
Power Mgmt
power.ti.com
Space, Avionics and Defense
www.ti.com/space-avionics-defense
Microcontrollers
microcontroller.ti.com
Video and Imaging
www.ti.com/video
RFID
www.ti-rfid.com
OMAP Applications Processors
www.ti.com/omap
TI E2E Community
e2e.ti.com
Wireless Connectivity
www.ti.com/wirelessconnectivity
Mailing Address: Texas Instruments, Post Office Box 655303, Dallas, Texas 75265
Copyright © 2015, Texas Instruments Incorporated